Characterization of chaotic mixing effects in hydrometallurgical leaching process based on deep learning

Abstract

Traditional stirring methods in the wet metallurgy leaching process suffer from low efficiency, high consumption, and low output, leading to increased production costs and energy consumption. Therefore, this study evaluates reactor performance using deep learning and introduces variable-speed stirring to enhance laminar mixing and reduce stirring reactor power consumption. An S-Type acceleration and deceleration control algorithm is constructed to ensure that stepper motors do not experience step loss, stalling, or overshoot when the frequency changes abruptly. A deep learning tracking model based on dual cameras is established to dynamically track tracer particles inside the stirring reactor, and a Euclidean distance evaluation method is proposed to characterize and evaluate the mixing performance of the stirring reactor. Experimental results demonstrate that the use of complex function variable-speed stirring and shortening the variable-speed cycle both contribute to improving mixing efficiency. Under a variable-speed cycle of 5 s, chaotic speed increases mixing efficiency by 53.1 % compared to constant speed. This study provides a theoretical basis for optimizing the wet metallurgy leaching process.